-
How Much You Ate? Food Portion Estimation on Spoons
Authors:
Aaryam Sharma,
Chris Czarnecki,
Yuhao Chen,
Pengcheng Xi,
Linlin Xu,
Alexander Wong
Abstract:
Monitoring dietary intake is a crucial aspect of promoting healthy living. In recent years, advances in computer vision technology have facilitated dietary intake monitoring through the use of images and depth cameras. However, the current state-of-the-art image-based food portion estimation algorithms assume that users take images of their meals one or two times, which can be inconvenient and fai…
▽ More
Monitoring dietary intake is a crucial aspect of promoting healthy living. In recent years, advances in computer vision technology have facilitated dietary intake monitoring through the use of images and depth cameras. However, the current state-of-the-art image-based food portion estimation algorithms assume that users take images of their meals one or two times, which can be inconvenient and fail to capture food items that are not visible from a top-down perspective, such as ingredients submerged in a stew. To address these limitations, we introduce an innovative solution that utilizes stationary user-facing cameras to track food items on utensils, not requiring any change of camera perspective after installation. The shallow depth of utensils provides a more favorable angle for capturing food items, and tracking them on the utensil's surface offers a significantly more accurate estimation of dietary intake without the need for post-meal image capture. The system is reliable for estimation of nutritional content of liquid-solid heterogeneous mixtures such as soups and stews. Through a series of experiments, we demonstrate the exceptional potential of our method as a non-invasive, user-friendly, and highly accurate dietary intake monitoring tool.
△ Less
Submitted 11 May, 2024;
originally announced May 2024.
-
In The Wild Ellipse Parameter Estimation for Circular Dining Plates and Bowls
Authors:
Akil Pathiranage,
Chris Czarnecki,
Yuhao Chen,
Pengcheng Xi,
Linlin Xu,
Alexander Wong
Abstract:
Ellipse estimation is an important topic in food image processing because it can be leveraged to parameterize plates and bowls, which in turn can be used to estimate camera view angles and food portion sizes. Automatically detecting the elliptical rim of plates and bowls and estimating their ellipse parameters for data "in-the-wild" is challenging: diverse camera angles and plate shapes could have…
▽ More
Ellipse estimation is an important topic in food image processing because it can be leveraged to parameterize plates and bowls, which in turn can be used to estimate camera view angles and food portion sizes. Automatically detecting the elliptical rim of plates and bowls and estimating their ellipse parameters for data "in-the-wild" is challenging: diverse camera angles and plate shapes could have been used for capture, noisy background, multiple non-uniform plates and bowls in the image could be present. Recent advancements in foundational models offer promising capabilities for zero-shot semantic understanding and object segmentation. However, the output mask boundaries for plates and bowls generated by these models often lack consistency and precision compared to traditional ellipse fitting methods. In this paper, we combine ellipse fitting with semantic information extracted by zero-shot foundational models and propose WildEllipseFit, a method to detect and estimate the elliptical rim for plate and bowl. Evaluation on the proposed Yummly-ellipse dataset demonstrates its efficacy and zero-shot capability in real-world scenarios.
△ Less
Submitted 11 May, 2024;
originally announced May 2024.
-
Double-Condensing Attention Condenser: Leveraging Attention in Deep Learning to Detect Skin Cancer from Skin Lesion Images
Authors:
Chi-en Amy Tai,
Elizabeth Janes,
Chris Czarnecki,
Alexander Wong
Abstract:
Skin cancer is the most common type of cancer in the United States and is estimated to affect one in five Americans. Recent advances have demonstrated strong performance on skin cancer detection, as exemplified by state of the art performance in the SIIM-ISIC Melanoma Classification Challenge; however these solutions leverage ensembles of complex deep neural architectures requiring immense storage…
▽ More
Skin cancer is the most common type of cancer in the United States and is estimated to affect one in five Americans. Recent advances have demonstrated strong performance on skin cancer detection, as exemplified by state of the art performance in the SIIM-ISIC Melanoma Classification Challenge; however these solutions leverage ensembles of complex deep neural architectures requiring immense storage and compute costs, and therefore may not be tractable. A recent movement for TinyML applications is integrating Double-Condensing Attention Condensers (DC-AC) into a self-attention neural network backbone architecture to allow for faster and more efficient computation. This paper explores leveraging an efficient self-attention structure to detect skin cancer in skin lesion images and introduces a deep neural network design with DC-AC customized for skin cancer detection from skin lesion images. The final model is publicly available as a part of a global open-source initiative dedicated to accelerating advancement in machine learning to aid clinicians in the fight against cancer.
△ Less
Submitted 20 November, 2023;
originally announced November 2023.
